MOF, Alm.del - 2015-16 - Endeligt svar på spørgsmål 866: Spm. om professor Stiig Markager materiale om danske kvælstoftilførslers indflydelse på vandets klarhed i Kattegat-Bælthavet m.m. fremlagt under høringen den 23. februar 2016, til miljø- og fødevareministeren

Miljø- og Fødevareudvalget 2015-16
MOF Alm.del
Offentligt

Limnol. Oceanogr.,

59(5), 2014, 1679–1690

2014, by the Association for the Sciences of Limnology and Oceanography, Inc.

doi:10.4319/lo.2014.59.5.1679

Changes in the vertical distribution of primary production in response to

land-based nitrogen loading

Maren Moltke Lyngsgaard,

1,2,*

Stiig Markager,

and Katherine Richardson

Department

Center

of Bioscience, Aarhus University, Roskilde, Denmark

for Macroecology, Evolution and Climate, Danish Natural History Museum, University of Copenhagen, Copenhagen, Denmark

Abstract

Anthropogenic nitrogen (N)-loading has decreased significantly in the Baltic Sea Transition Zone over the past

two decades. We show that the vertical distribution of primary production (PP) changed as a function of land-

based N-loading using 1385 water column photosynthesis estimates, in which photosynthetic parameters were

determined both in the surface water layer and in the pycnocline-bottom layer (PBL) at six stations near the

Danish coast between 1998 and 2012. Total annual PP and surface layer PP (SPP) correlate positively with land-

based N-loading from Denmark (p

0.003). The percentage of annual PP occurring in the PBL (denoted as deep

primary production, DPP) varied annually between 6% and 30% (mean

17%). The absolute magnitude of the

DPP, as well as its relative proportion of total water column PP, correlates negatively with N-loading (p

0.009

and

0.0003, respectively). Thus, SPP decreases in response to decreased N-loading, while DPP increases.

Land-based N-loadings also correlate positively with the light attenuation coefficient (R

0.39,

0.05), which

may in part explain the response in DPP to changes in N-loading. DPP occurs in active phytoplankton

communities acclimated and/or adapted to low light and producing oxygen in the PBL water.

Primary production (PP) is an important factor in

structuring marine ecosystems, and changes in PP in

response to increased nutrient loading have been identified

as being responsible for symptoms of eutrophication in

coastal marine systems (Nixon 1995; Cloern 2001; Smith

2003). After recognizing the relationship between anthro-

pogenic nutrient loading and eutrophication, many coun-

tries have initiated programs to reduce nutrient enrichment

(Boesch 2002) with the expectation that marine PP will

respond to a reduction in land-based nutrient loading. In

some areas, enough data have now been collected to allow

researchers to examine the extent to which this expectation

has been realized.

The Baltic Sea Transition Zone (BSTZ), which com-

prises the Kattegat and the Belt Seas and thus forms the

connection between the Baltic Sea and the Skagerrak, is

such an area. It is a shallow, stratified, and temperate

marine system with dynamic hydrography. Surface salinity

varies in the region from 10 to 14 in the southern part of the

region and from 20 to 25 in the northern Kattegat

(Gustafsson 2000). Bottom water salinity generally varies

between 32 and 34, and an essentially permanent pycno-

cline is present. The area can be characterized as a frontal

system in which the low-saline surface water from the Baltic

Sea mixes with the more saline waters coming from the

Skagerrak. The system demonstrates clear estuarine circu-

lation where water transport is mainly driven by the water

level difference between the Arkona Sea and the Northern

Kattegat and, ultimately, by the freshwater surplus to the

Baltic Sea of 559 km

(Savchuk 2005). Mixing with

surface water, especially in the Little and Great Belts (see

Fig. 1), ventilates the bottom water of the BSTZ (Bendtsen

et al. 2009).

* Corresponding author: [email protected]

The seasonal distribution of PP here is typical for

temperate coastal waters: Elevated production occurs in

association with the spring phytoplankton bloom, but peak

PP occurs during the summer months (Petersen and Hjorth

2010), as is also seen in other temperate estuarine regions

such as the Chesapeake Bay (Kemp and Boynton 1984).

Winter PP is low and limited by light availability.

Anoxia and hypoxia events in the BSTZ became more

widespread over the 20th century (Conley et al. 2007). As

these were believed to be a consequence of increases in

anthropogenic nutrient loading, legislation was established

in the late 1980s to control and reduce land-based nutrient

loading (Conley et al. 2002). Following the establishment of

this legislation, Denmark has maintained an extensive

marine monitoring program. The data set resulting from

this monitoring program provides a unique resource for

identifying relationships between PP and changing nutrient

loadings, which may be of potential relevance in under-

standing eutrophication responses in other coastal areas.

The purpose of this study was to examine these

monitoring data for evidence of a response in annual PP

in this region to changing nitrogen conditions. Because it is

well known that the BSTZ, as well as other seasonally or

semi-permanently stratified areas (Richardson et al. 2003;

Lehrter et al. 2009; Strom et al. 2010), are characterized by

a significant amount of annual PP taking place in

association with subsurface phytoplankton peaks (Cullen

1982; Richardson and Christoffersen 1991; Karlson et al.

1996), we chose not only to examine total water column PP

but also the seasonal and interannual variability in the

vertical distribution of PP. We hypothesized that PP

occurring in the pycnocline-bottom layer (referred to as

deep primary production, or DPP) may respond differently

than PP in the surface waters (SPP) to changes in land-

based nitrogen (N)-loading.

1679

MOF, Alm.del - 2015-16 - Endeligt svar på spørgsmål 866: Spm. om professor Stiig Markager materiale om danske kvælstoftilførslers indflydelse på vandets klarhed i Kattegat-Bælthavet m.m. fremlagt under høringen den 23. februar 2016, til miljø- og fødevareministeren

1680

Lyngsgaard et al.

for the discrete depths at which sampling had been

conducted.

Data for total nitrogen (TN) loadings from Denmark to

the BSTZ were also taken from the MADS database. The

values are based on a three-dimensional Mike She (Windolf

et al. 2011) groundwater resource model validated on

measured TN concentrations in Danish streams. The

hydrological area chosen for this study covered the inner

Danish waters and ranged from the Kattegat Sea in the

north to the Belt Seas in the south.

The daily surface photosynthetic active radiation

(SPAR) was determined by averaging measurements made

continuously at several different localities in Denmark

within approximately 30–150 km of the sample locations.

Daily insolation was estimated using average SPAR

(between the different locations) calculated for every half

hour, and these average values were used to calculate the

24 h depth-integrated PP for all six stations. This data set is

not publically available.

Only data from 1998 through 2012 were used. This

period was chosen as the protocol for measuring

photosynthesis parameters was altered in 1998 so that

parameters were measured at two depths rather than one.

These two depths were an integrated ‘surface’ sample from

0 to 10 m (unless visual examination of the density profile

indicated a major density difference or pycnocline within

this depth range) collected using a hose inserted to the

desired depth or with Niskin bottles at 1, 5, and 10 m in

depth (or to the depth of the pycnocline) and the depth of

the deep chlorophyll maximum, defined as the depth with

the highest chlorophyll fluorescence and where fluores-

cence was greater than twice the average for the surface

layer (sample collected using Niskin bottle). Chl

concentrations were also determined in these two samples.

The protocol for the monitoring program prescribed that

if no deep chlorophyll maximum was encountered, the

second sample was taken at the depth of 2% surface light

penetration.

In addition to the two depths described above, Chl

was

analyzed at standard depths (every 5 m) throughout the

water column starting at 1 m below the surface and ending

1 m above the sediment. Chl

determinations were made

by filtering samples onto Whatman GF/F or GF 75

Advantec filters. Filters were extracted in ethanol (96%)

for 6–20 h and samples analyzed spectrophotometrically

according to the method described by Strickland and

Parsons (1972) and modified by Danish Standards (1986).

Photosynthetic carbon assimilation was estimated based

on the carbon-14 method modified (Markager 1998) after

Steemann Nielsen (1952). Photosynthesis vs. irradiance

curves (P vs. E curves) were calculated from incubations

made under artificial light (Osram HQI-T or high-pressure

halogen lamps), where the samples were incubated at seven

different light intensities for 2 h with metal grids providing

approximately 35% light attenuation between each bottle.

Thereafter, the samples were filtered (GF/F or GF 75

Advantec filters) and the carbon incorporation stopped

with acid (200

0.1 mol L

HCl). The amount of

incorporated carbon-14 in the phytoplankton was determined

by liquid scintillation counting resulting in disintegrations per

Fig. 1. The Baltic Sea Transition Zone (BSTZ) and location

of the six stations.

Methods

Danish National Aquatic Monitoring and Assessment

Program—Data

from the 1998–2012 time period were

obtained from the database of the Danish National

Aquatic Monitoring and Assessment Program (MADS;

Conley et al. 2002). This database is publically available

(http://www2.dmu.dk/1_viden/2_Miljoe-tilstand/3_vand/4_

mads_ny/default_en.asp) and contains physical, chemical,

and biological data collected in BSTZ waters since the

1980s. Sampling and measurements were done by different

laboratories but followed a common set of technical

guidelines and procedures (Kaas and Markager 1998), and

inter-comparisons of results were frequently carried out by

the former Danish National Environmental Research

Institute (NERI), now Aarhus University. For the param-

eters used in this study, the inter-comparisons were

supervised by one of the authors (S.M.).

Six study stations were selected based on (1) a minimum

depth of 10 m and (2) the amount of data available. Study

station depths vary from 14 m (Aalborg Bight) to 51 m in

The Sound (Fig. 1). The four stations of intermediate

depths are located in Aarhus Bight, the Little Belt, and the

Great Belt (two stations). For each station, the following

data were extracted: conductivity, temperature, and depth

(CTD) profile data with 0.2 m resolution for the vertical

distribution of temperature, salinity, chlorophyll fluores-

cence, and photosynthetically active radiation (PAR; 4p

sensor). In addition, measurements of chlorophyll

(Chl

and photosynthetic parameters (see below) were extracted

Baltic primary production and N-loading

1681

Fig. 2. Example of how parameters were derived for the primary production estimates: Profiles from 19 July 2010 in Aarhus Bight

of (a) density (r) and density difference m

(Dr

, kg m

). The starting depth of the pycnocline-bottom layer (PBL) was defined as

the first depth at which

Dr Dz

1 kg m

and, in the example here, was found at 6 m. (b) Interpolation of the chlorophyll-specific

photosynthetic parameters,

and P

. Actual determinations of these parameters at 1 and 15 m are indicated by black dots. (c) Primary

max

production, irradiance, and chlorophyll

Irradiance is the sum over 24 h in 20 cm depth intervals.

minute (DPM). The equation used to calculate the carbon

assimilation in each sample was as follows:

Tot CO

DPM

tot

{DPM

dark

Þ|1:05|60

(DPM

start

|Light

minutes)

ð1Þ

where P is the measured production (mg C m

), Tot CO

is the total dissolved inorganic carbon concentration deter-

mined by Gran-titration, and Light minutes is the time (in

minutes) the phytoplankton have been exposed to light. The

C isotope discrimination factor is 1.05, and the value 60 is

used to calculate from minutes to hours. Light attenuation at

each station was determined by estimating the diffuse light

attenuation coefficient (K

) from the CTD profile using a

deck sensor as a reference.

Data analyses—The division of the water column into

two layers:

Density (r) was calculated from salinity and

temperature profiles (Fofonoff 1985). These density pro-

files were used to calculate the depth separating the surface

from the layers below (i.e., the pycnocline-bottom layer

[PBL]). This was defined as being at the first vertical

density gradient of

1 kg m

(see Fig. 2a for an

example). For each day with a distinct pycnocline (during

which this density criteria was met), every depth in the

water column was assigned as being in one of two layers

(i.e., in the surface layer or in the PBL). This gave the

possibility of considering the PP occurring in the surface

layer (surface primary production, SPP) and below the

surface (i.e., in the PBL [deep primary production, DPP]),

respectively. When the density criteria of

Dr/Dz

1 kg m

was not met in the water column, the total water column

PP was considered as being SPP.

Two hundred fifty-six visual inspections of density

profiles were made to confirm that the density criterion

was effective at determining the starting depth of the PBL.

The depths found by visual inspection matched the depths

found by the criterion in 90% of the cases. The density

criterion chosen (i.e., 1 kg m

) overestimated the starting

depth of the PBL (i.e., identified a deeper [1–2 m] depth

than identified visually) in the remaining 10%. Thus, our

estimate of the fraction of DPP as a percentage of total

water column PP (see below) is conservative.

Estimating Chl

from fluorescence—Fluorescence

per

unit of Chl

changed systematically with depth in the data

set. Therefore, a fluorescence factor (F

Chl

F/[Chl]) was

calculated whereby the chlorophyll concentration recorded

in the discrete sample (Chl) and the fluorometer measure-

ment from the corresponding depth (F) were related.

Values for F

Chl

between sampling depths were assigned by

linear interpolation. This was done with 0.2 m resolution

and the resulting profile of F

Chl

(z) used to estimate a

continuous Chl

profile (Chl(z)

F(z)/F

Chl

(z)) (see Fig. 2c

for an example of an estimated Chl

profile).

Estimating PP through the water column—To

estimate

water column PP, chlorophyll-specific photosynthesis rates

were calculated from the P vs. E curves and Chl

1682

Lyngsgaard et al.

Table 1.

Number of depth-integrated primary production measurements per year at the six sampling locations. In all, there were

1385 primary production measurements from year 1998 to 2012.

1998 1999

Aalborg Bight

Aarhus Bight

Little Belt

Great Belt 1

Great Belt 2

The Sound

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

concentrations for the two samples, according to the method

of Markager et al. (1999). A light matrix representing the

light intensity at 0.2 m intervals throughout the water column

and hourly intervals over the entire day was constructed

using the attenuation coefficient and the surface light. This

light matrix and the photosynthetic characteristics derived

from the P vs. E curves (i.e.,

max

, and offset (c) for the two

sampling depths normalized to Chl

concentrations [a

, and c

]) were combined to estimate daily PP at 0.2 m

max

intervals throughout the water column for all stations except

the Great Belt 1 Sta., at which only P vs. E curve parameters

from the surface layer were available. The P vs. E parameters

were obtained by a non-linear fitting procedure (Statistical

Analysis System, SAS, version 9.4) on the carbon uptake

from Eq. 2 divided by the chlorophyll concentration. The P

vs. E model was taken from Webb et al. (1974), modified by

the inclusion of an offset (c), thus:

ð2Þ

max

1{exp

{

max

where

is the slope for the light-limited part of the P vs. E

curve (g C g

Chl

mol

), and P

the light-saturated

max

photosynthetic rate (g C g

Chl

). The offset (c) was

included in order to avoid a bias in

as a result of the

fact that

C-uptake rates are always positive.

See

also

Markager et al. (1999) for details about the curve fitting

procedure.

It was assumed that the P vs. E parameters from the

surface sample represented an average value for the entire

surface layer, and these values were therefore extracted from

the surface to the starting depth of the PBL (see Fig. 2 for an

example). The reason for using the same P vs. E parameter

values throughout the surface layer was that turbulence

within this layer was assumed to be too high to allow further

depth-specific photo adaptation (Lewis et al. 1984). Turbu-

lence in the PBL was assumed to be lower, thus allowing more

variation in the P vs. E parameter values with depth.

Therefore, these were interpolated from the starting depth

of the PBL (5 surface values) to the second sampling depth.

From the second sampling depth and downward photosyn-

thetic parameters were assumed to be constant (Fig. 2b). The

volumetric PP at each 0.2 m interval, P

vol

, (mg C m

)

was calculated from Eq. 3 with the following parameters:

vol

[Chl],

vol

[Chl], and c

vol

[Chl].

max

PAR

vol

1{exp

vol vol

ð3Þ

vol

max

where PAR is the product of surface irradiance and the PAR

fraction left at a given depth [PAR fraction

reflection)

exp (2K

(depth

0.05))]. The surface reflection was

assumed to be 6%.

In this manner, estimates of total water column PP as

well as the contribution made by DPP to the total were

made for the 1385 d of sampling. Data availability varied

between stations and, for each of the six stations, ranged

from 3 to 14 yr, resulting in a final time series of 15 yr

(Table 1). The sampling frequency at each station ranged

from 12 to 49 measurements per year.

The average annual values of PP were calculated for all

six stations. Long-term average monthly values were

calculated for each station (including all available years),

and these were used to fill months during which there were

no measurements. If data for entire years were missing,

these years were excluded from the analysis. This procedure

allowed us to examine the temporal development in PP

without the influence of missing values in the data set. We

used the same approach with respect to missing values for

the P vs. E parameters and the light attenuation coefficient

) when analyzing their interannual and seasonal

variations.

Statistical analyses—Multiple

linear regressions were

constructed to examine the degree to which the estimated

PP could be explained by irradiance and N-loading.

Annual PP, annual DPP (g C m

), DPP in

percentage of total, annual land-based N-loading to the

BSTZ from Denmark (10

kg of nitrogen), irradiance, and

the light attenuation coefficient (K

) were normalized to

their mean values over the study period. In this way the

unit of the coefficients for the regressions performed

becomes the percentage change in dependent variable

(PP, SPP, DPP, DPP in

and K

) per percentage change

in forcing variables (N-loading and surface irradiance).

The annual land-based N-loading was calculated on a

monthly basis and summed over a period of consecutive

months. When relating PP parameters to N-loading, we

first regressed the relationships between the parameters for

a given calendar year against total N-loading in the same

calendar year. In recognition of the fact that loading might

affect PP differently over the season and that there likely

will be a lag between loading and PP response, we then

tested different periods for loading against the annual PP

parameters. A systematic approach was used in choosing

the periods to be tested. We started with 24 months

covering the whole calendar year before and the year of the

given annual PP. We then excluded and included months so

Baltic primary production and N-loading

Table 2.

Statistics for multiple linear regressions between primary production and the diffuse light attenuation coefficient (K

) vs. land-based N-loading from Denmark

and surface irradiance. The time period extends from 1998 to 2012. In addition, the temporal development in K

is analyzed from 1990 to 2012. Both dependent and

independent variables are scaled to the mean value

100; thus, the unit of the coefficients becomes

change in dependent variable per

change in the independent variable

(i.e.,

change in PP/% change in N-loading). Coefficients and intercepts are given

standard errors, and values in brackets are

p-values

for significance. Finally, each

dependent parameter is tested against time (annual change). Significant changes (p

0.05) are indicated in bold type.

1683

0.9960.64(0.15)

4.4±1.96(0.0006)

1.48±0.61(0.035)

3.2±0.66(0.0004)

Not significant

Coefficient (C

)

for surface

irradiance

that all possible periods having a minimum length of

2 months were tested. In all, 276 different periods of N-

loading were tested against annual PP, SPP, and DPP. The

periods giving the highest explanatory power (R

) are

shown in Table 2.

The irradiance values used in the statistical analyses

represent the actual SPAR at the surface from April to

December as in all cases this period was that with highest

predictive value. A similar procedure to the one described

above was applied when the N-loading was considered in

relation to K

An analysis of the temporal development in the depth

separating the surface and PBL was carried out on

normalized depths (i.e., to the average depth of the PBL

for the individual station). When the normalized starting

depth of PBL was calculated for each station, a linear

regression was used on the normalized values as a function

of years per station and then for an average of the six

stations.

0.69

0.72

0.56

0.80

0.27

0.39

260±70

2276±98

292667

2144±68

89±5

90±4

Intercept

22.0(0.09)

3.6(0.08)

21.0(0.33)

3.6(0.026)

20.2(0.66)

20.9(0.017)

Annual

change,

Period for sum

of surface

irradiance

Results

Apr–Dec

—

Surface primary production

Deep primary production

Total water column production

Deep primary production in

of total

Diffuse attenuation coefficient (K

; 1998–2012)

Diffuse attenuation coefficient (K

; 1990–2012)

* Indicates months in the year before the calendar year for which primary production is calculated.

Dependent variable

Total and vertical distribution of PP—The

mean annual

PP from 1998 to 2012 for the six stations was 189, with

an interannual variation in the annual mean of

17.2

g C m

. The lowest production estimates were found

in Aalborg Bight, Aarhus Bight, and The Sound (155–

175 g C m

), while the production in Great and Little

Belts was higher (209–226 g C m

; Fig. 3). The

higher values at these stations likely reflect nutrient inputs

from pycnocline-bottom to surface waters through more

intense mixing in the narrow straits (Lund-Hansen et al.

2008). The highest water column PP was found during

summer, and the seasonal pattern in production largely

followed irradiance and temperature (data not shown).

When all stations in the entire study period are

considered, the DPP, on average, contributed 17% to the

annual PP. The contribution of the DPP to the total water

column production varied from 6% in the Little Belt to

30% in Aalborg Bight (Fig. 3). The rest of the stations

exhibited contributions of DPP to annual production of

between 13% and 23% (Fig. 3). The highest average

monthly contribution from DPP to total was observed in

Aalborg Bight in May, where the DPP contributed 48% to

total water column production (Fig. 3c). DPP was low or,

in some cases, nonexistent from October to February, when

the phytoplankton community was well mixed throughout

the water column and surface irradiance was too low to

support production in the PBL (Fig. 3). Overall, the

importance of DPP was slightly higher in summer from

April to August (8–40% of summer production) than for

the year as a whole.

Photosynthetic parameters—The

seasonal pattern of the

light intensity at which photosynthesis initially is saturated,

max

/a,

mmol

photons m

) is shown in Fig. 4a as

the average for all six stations. Samples from the surface

layer show increasing I

values from January (68

mmol

photons m

) to a peak in August (128

mmol

photons

), followed by a decline to 78

mmol

photons m

Period for sum

of N-loading

Coefficient (C

) for

N-loading from

Denmark

Feb–Sep

Mar*–Feb

Feb–Sep

Mar*–Feb

Nov*–Mar

0.61±0.12(0.0002)

20.66±0.21(0.0090)

0.43±0.11(0.0025)

20.73±0.14(0.0003)

0.10±0.05(0.0495)

0.09±0.03(0.0121)

1684

Lyngsgaard et al.

Fig. 3. Monthly average values for total water column primary production (PP) and deep PP (DPP) for six different stations in the

BSTZ (1998–2012). The annual PP and the contribution (in

from DPP to PP are given in the upper right corner of each graph.

in December. For the deep samples, the seasonal variation

was smaller. I

increased from 62

mmol

photons m

May to 89

mmol

photons m

in November and then

dropped to 55

mmol

photons m

in January. A

t-test

showed significantly higher average monthly I

values in

the surface layer than in the PBL (t-test,

3.56, degrees

of freedom [df]

22,

0.01,

24). In addition to the

characteristic seasonal pattern seen for the I

values from

the surface layer, a significant and positive relationship was

found between I

values from the surface layer and the

daily surface irradiance (r

0.28,

0.0001,

1080).

Thus, the I

data indicate that the phytoplankton

communities in and below the surface layer are acclimated

to the prevailing light levels they encounter.

Not surprisingly, this conclusion is also supported by the

general patterns observed for maximum chlorophyll-specific

photosynthesis rate (P

), as this parameter controls a large

max

part of the variation in I

. Average monthly values

(including data from 15 yr) for P

showed significantly

max

higher values for phytoplankton populations found in the

surface layer (P

3.6

1.4 g C g

Chl

) than in

max

the PBL (P

2.3

0.4 g C g

Chl h

;

t-test, t

3.06,

max

22,

0.01,

24). An exception to this general

pattern was observed for the Sound, where P

values from

max

the two depths were similar.

Monthly averages of P

values in the surface layer for

max

all six stations for the 1998–2012 period exhibited a

seasonal pattern in which P

increased from 2.4 g C g

max

Chl h

in March to 6.4 g C g

Chl h

in August and

then decreased to 2.2 g C g

Chl h

in December (Fig. 4).

max

values from surface and PBL waters were not

significantly different (t-test,

1.83, df

734,

0.07,

736) between the months of December and

March. This can also be seen from the average monthly

values (Fig. 4). During this period, the water column was

often well mixed and the solar irradiance so low that

photosynthesis was likely light limited most of the time.

The general pattern in the chlorophyll-specific initial

slope (a

) of the P vs. E curves exhibited no significant

difference for phytoplankton populations found in the

surface layer and those found in the PBL (Fig. 4c). In

addition, the variability, expressed as the coefficient of

variance (CV), of

was lower than that of P

(CV

max

0.16 compared to CV for P

0.42). This is to be

max

expected, as

is equal to absorption times quantum yield

(Markager and Vincent 2001), and light absorption is

tightly coupled to the chlorophyll content in the cells

(Markager 1993). A weak seasonal pattern in

values was

detected that resembled that found for I

and P

. The

max

values increased from 10.2 g C g

Chl mol photons

in March to the peak value in August (14.2 g C g

Chl

mol

), and decreased thereafter to 8.2 g C g

Chl

mol

in December. Higher values were found in

January and February than in March and April.

PP and land-based N-loading—The

concentration of

nitrogen in land runoff from Denmark (mg N L

)

decreased significantly between 1998 and 2012 (linear

regression:

0.0001,

0.84,

15; Fig. 5),

primarily as a result of the establishment of water quality

regulation (Windolf et al. 2012). Although there were large

fluctuations in the absolute loading due to interannual

variation in runoff, the absolute delivery of nitrogen from

Denmark (10

kg N) also showed a significant decrease

over the study period (linear regression:

0.016,

0.37,

15;

see

Fig. 5).

Total PP, SPP, and DPP, as well as the ratio of SPP

and DPP to total PP, were all significantly correlated to

Baltic primary production and N-loading

1685

Fig. 5. Land-based annual N-loading (right y-axis) and land-

based annual N-loading normalized to runoff (left y-axis) from

1998 to 2012. The numbers are based on average values of Danish

N-loading to the study area.

Fig. 4. Seasonal variation of chlorophyll

a–specific

P vs. E

parameters: (a) the light intensity at which photosynthesis is

initially saturated, I

; (b) maximum chlorophyll-specific photo-

synthesis rate, P

; and (c) the initial slope of the P vs. E curve,

max

. Parameters are shown for 1–10 m (depth 1) and for depths in

the pycnocline-bottom layer (depth 2). The parameters are

average values of six stations in the Baltic Sea Transition Zone

(1998–2012), and the error bars indicate the variation in average

monthly values among stations.

land-based N-loading (Table 2). However, while SPP was

positively correlated to N-loading (see Fig. 6), the opposite

was true for DPP.

The relationship between SPP and N-loading can be seen

in Fig. 6, where both SPP and N-loading are calculated for

the calendar year. The relationship is significant, with a

value

0.01 and a coefficient of 0.47% per

change in N-

loading. Thus, when N-loading changed by 1%, the SPP

changed by 0.47% in the same direction. It is, however,

doubtful that N-loading calculated over the calendar year is

the best descriptor in terms of relating N-loading to PP, as

N-loading late in the calendar year obviously cannot affect

the period of peak production during the same year.

Instead, it can be argued that nitrogen has a residence time

in the system. This would argue for including N-loading

from the previous year when considering PP in a given year,

especially given the fact that phytoplankton growth is only

believed to be N-limited during the summer period in the

region (Conley 1999). Following this line of reasoning, the

N-loading during the summer might be most important in

controlling PP. We therefore tested a number of different

loading periods extending back to January of the year prior

to the test year in relation to the PP variables.

A change in N-loading was the best predictor of changes

in SPP when the 8 months from February to September of

the study year were included (coefficient of 0.61;

0.001;

Table 2). Surface irradiance was also positively related to

SPP, but the coefficient was not significant. Figure 7 shows

the effect on the

value (explained variance) by varying

the timing of an 8 month loading period. All 8 month

periods with a midpoint from 01 November in the year

prior to the study year to 01 July of the study year yielded

significant (p

0.05) coefficients for the correlation

between SPP and N-loading. The value of the coefficient

varied from 0.42 to 0.61. This suggests that the N-loading

from the previous winter and during the period in which

phytoplankton activity is greatest are the most important

drivers of SPP. Calculating the N-loading over periods with

different lengths gave similar results.

1686

Lyngsgaard et al.

Fig. 6. Changes in annual primary production in the surface

layer as a function of changes in N-loading. Change in each

parameter is expressed as a percentage of the average value for the

study period (1998–2012) as a whole. A significant linear

relationship between the two parameters (p

0.01 and

0.41) is noted on the graph with a dashed line.

Fig. 7. An example of the relationships between periods over

which the N-loading is calculated as well as the resulting

values

and coefficients. The month is the midpoint of the period (an

asterisk indicates the year before the calendar year during which

primary production is calculated), and the period from Table 2 is

indicated in bold. The horizontal dashed line indicates when

the

p-value

is below 0.05. The example shown here is for

surface primary production vs. N-loading calculated over an

8 month period.

The DPP was strongly and positively correlated with

surface irradiance (coefficient

4.4%/%,

0.001) but

negatively related to the N-loading (Table 2). In this case,

the most significant relationship was found when N-

loading was calculated over a 12 month period starting in

March of the year prior to the study year to February of

the study year (midpoint between August and September of

the year prior to the study year). The coefficient to N-

loading was negative for all periods but was only significant

for 12 month periods with midpoints between August and

October of the year prior to the study year (Table 2), which

means that changes in DPP were best predicted by a period

of N-loadings occurring prior to the actual DPP.

Total PP was also significantly and positively related to

N-loading (summed from February to September of the

study year; i.e., similar to the case for SPP). However, the

relationship was weaker than the relationship between

loading and SPP. This can easily be explained as the total

water column PP is the sum of SPP and DPP and these two

layers exhibited opposite responses to N-loading.

The most significant model relating PP parameters to N-

loading was that describing the fraction of DPP (as a

percentage of total water column production) as a function

of N-loading. This indicates that N-loading potentially

redistributes PP in the water column so it shifts from the

surface layer to the PBL as loadings are reduced. This result

cannot be attributed to changes in the depth at which the

surface and PBL diverge as there was no significant temporal

development in the starting depth of the PBL when the

stations were analyzed by linear regression separately (p

0.05) as well as when they were averaged into one value

(normalized to the average depth for the study period as a

whole) per year (p

0.05). A possible explanation for the

identified negative relationship between DPP and N-loading

is the positive relationship found between N-loading and

light attenuation in the water column. We therefore also

tested the relationship between N-loading and the diffuse

attenuation coefficient (K

). A significant positive relation-

ship was found with a coefficient of 0.10% in K

per percent

change in N-loading (Table 2). K

data were available from

1990. Using this longer time series gave the same coefficient

but with a lower

p-value

(Table 2). There was a positive

relationship between the interannual variation (relative to

the long-term average) in K

and surface layer chlorophyll

0.30,

0.05,

15). Thus, the ‘extra’ PP stimulated

by a higher N-loading increases the chlorophyll concentra-

tion in the surface layer, contributing to the drop in water

transparency.

Temporal development in biological and physical param-

eters—The

temporal trends of the studied parameters in

response to changes in land-based N-loading are masked by

the large interannual variability in runoff. Nevertheless, we

also regressed all of the parameters studied against year

(Table 2, last column). While the trends expected in light of

the overall reduction in N-loadings were found, two more

parameters showed a significant temporal development.

The K

decreased over time (for the time period from 1990

to 2012) and the fraction of DPP to total PP showed an

increase from 1998 to 2012 (see Table 2). No temporal

trends were found for PP, SPP, or DPP (as absolute carbon

values). Thus, over the 1998–2012 period, there has been a

significant vertical redistribution of PP in the BSTZ,

whereby PP has moved from the surface waters to the

PBL and water clarity has increased.

Discussion

PP in relation to land-based N-loading—This

study

demonstrates a redistribution of PP in the water column

Baltic primary production and N-loading

with decreasing land-based N-loading from Denmark,

whereby the proportion of PP taking place beneath the

surface layer increases with decreasing loading. This

finding is potentially important for the management of

eutrophication induced by human activities. From our

general understanding of eutrophication (Cloern 2001),

such a redistribution of PP in response to anthropogenic

eutrophication might have been predicted. However, to our

knowledge, this is the first time it has been demonstrated

for a coastal ecosystem on a decadal timescale. That it has

not been documented earlier with field observations is

likely due to the fact that there are few locations with

substantial time series of data describing phytoplankton

photosynthetic activity over a period with concurrent

reduction in nutrient loading.

The coefficient of the regression relating SPP and N-

loading (Table 2) indicates that SPP changes 0.61% per

percent change in N-loading (using load period from

February to September of the study year). The intercept for

this regression (intercept

60%) indicates that more than

half of the production is based on regenerated N or N from

other sources rather than land-based loading (see below).

In the case of the DPP, a 1% increase in land-based N-

loading leads to a 0.66% decrease in DPP. An interesting

consequence of the fact that opposite responses to the

reduction of N-loading are seen in the two depth layers is

that the total water column production is not as sensitive to

changes in N-loading as are the two individual layers when

examined independently. Thus, total water column PP may

not be the ideal parameter to use in assessing system

responses to changes in nutrient loading.

That we see a greater shift in the distribution of PP in the

water column than in total PP is reminiscent of the pattern

observed for more shallow aquatic systems, in which an

increase in N-loading leads to a redistribution of PP from

benthic macrophytes to phytoplankton (i.e., higher loading

results in the production occurring closer to the surface;

Borum and Sand-Jensen 1996; Krause-Jensen et al. 2012).

As no temporal development in the starting depth of the

PBL was recorded, the most likely explanation for the

patterns observed in both cases must be that higher loading

leads to a reduction in water clarity. This is supported by

the relationship found in this study between the N-loading

and K

. That lower

values are found for K

(0.27–0.39)

compared to PP (0.56–0.80) in relation to nutrient loading

agrees well with earlier studies, in which it has been

demonstrated that the relationship between N-loading and

ecosystem effects becomes weaker when the response of the

parameter is not directly coupled to the loading being

examined (Timmermann et al. 2010; Carstensen et al.

2011).

The best fits between land-based N-loading and PP were

found when loading was summed over periods other than

the calendar year for which the PP was calculated. For SPP

and PP, the period best describing the relationship to N-

loading (judged on the basis of the value of

) was

February to September of the study year, but the winter

months prior to this period were also shown to be

important (Fig. 7). Thus, loading during the period when

PP is high has a large effect on the PP occurring. In

1687

contrast, both the DPP and K

values were better described

by N-loading occurring farther back in time (Table 2).

Assuming that the negative coefficient for N-loading vs.

DPP derives, ultimately, from the positive effect on light

attenuation, both observations would suggest the occur-

rence of a lag period from when the loadings reach the

marine system until the effects are observed.

Light attenuation is governed by several processes and

components in the water column (i.e., chlorophyll and

accumulated organic matter [CDOM or detritus]; Siegel

and Michaels 1996). CDOM is, in turn, derived partly from

the PP that has occurred earlier in the system. The positive

correlation found between chlorophyll in the surface layer

and K

indicates that light attenuation also is directly

affected by the amount of phytoplankton present in the

surface layer. However, as only 30% of the interannual

variation in K

in this study can be explained by variation

in surface layer chlorophyll, we attribute a major part of

the variation in K

occurring here to be related to the

amount of dissolved organic and particulate matter and the

light attenuation caused by water itself. It has earlier been

shown for this region that less than 30% (and often much

less) of the light attenuation is caused directly by

chlorophyll (Markager et al. 2010; Timmermann et al.

2010; Krause-Jensen et al. 2012). Thus, the interannual

relationship between N-loading and K

is directly affected

by the amount of chlorophyll in the surface layer and

further strengthened by the dissolved organic and partic-

ulate matter that comes with runoff water from land.

PP in the PBL—The

estimates of PP made with the

assumptions used here suggest that a significant proportion

of the PP in the study area occurs in the PBL. The average

contribution of this DPP to the annual PP for the six

stations ranged from 6% to 30%. This estimate agrees well

with previous findings in the region (Richardson and

Christoffersen 1991; Richardson et al. 2000, 2003).

Furthermore, the water column PP estimates indicate that

photosynthesis in the PBL is a consistent feature during the

period extending from April to October.

The chlorophyll-specific P vs. E parameters give support

to the conclusion derived from the PP estimates that a

photosynthetically active phytoplankton population is

found in the PBL. These parameters also suggest that the

phytoplankton population in the PBL is physiologically

distinct from populations in the surface layer. Both I

and

are lower in samples taken in the PBL than for

max

samples in the surface layer. This suggests that light is a

limiting factor in the PBL. The obvious adaptation and

acclimatization to low light would be higher

values.

However, chlorophyll-specific

values cannot be used to

detect this as

and chlorophyll content per cell will co-

vary. Chl

concentration within a single phytoplankton

cell increases during the process of acclimatization to lower

light intensities (Cullen 1982; Richardson et al. 1983).

Therefore, the lack of difference in

between depths may

be due to higher chlorophyll contents per cell for

phytoplankton from depth 2. Unfortunately, cell counts

were not made for the deep samples so we could not

calculate

per cell.

1688

Lyngsgaard et al.

here for the BSTZ has potentially important implications for

the production of oxygen in the PBL. Converting the

average DPP from this study (17% of annual production) to

oxygen equivalents (O

: CO

1) yields an oxygen

production in the PBL of about 89 g O

. Jørgensen

and Richardson (1996) examined data from the Southern

Kattegat and estimated the oxygen demand in the bottom

layer to be 202 g O

. Hansen and Bendtsen (2013)

argue that 28% of PP (, 54 g C m

based on data

from this study) is remineralized in the bottom waters. This

would be equivalent to an approximate oxygen demand of

144 g O

. Thus, the present study suggests that

oxygen produced by DPP in the BSTZ may compensate for a

considerable fraction of the oxygen demand in this layer.

Subsurface PP has also been shown to counteract,

although not eliminate, hypoxia in other areas where light

penetrates into the PBL (Lehrter et al. 2009; Murrell et al.

2009; Strom et al. 2010). Studies on the importance of

oxygen production below the surface layer for the

Louisiana continental shelf showed that up to 41% of the

bottom water oxygen demand may be resupplied by deep

PP, and when benthic production is included, this oxygen

contribution may be even higher (Dortch et al. 1994;

Lehrter et al. 2009).

Implications for understanding ecosystem effects of

anthropogenic nutrient loading—Nutrient

enrichment is well

known to influence both the magnitude of the production

of organic material in a system (Nixon 1995) and light

attenuation in the water column (Nielsen et al. 2002). Until

now, most studies of eutrophication in marine systems

have, therefore, focused on the influence of nutrient

enrichment on the total water column PP. The current

study suggests, however, that for the BSTZ, changes in

land-based N-loading influence the vertical distribution of

PP more than the total magnitude of PP. We see no reason

to believe that this should not be the case for other

stratified coastal systems as well.

This vertical redistribution of PP is potentially important

for the function of the system as a whole, as photosynthesis

occurring in the PBL introduces oxygen into the deeper

regions of the water column, regions that are not directly

able to exchange oxygen with the atmosphere and, thus,

ameliorate hypoxia. In addition, a change in the vertical

distribution of PP, as reported here, will likely have

consequences for the relative occurrence of different

phytoplankton species. This can potentially have conse-

quences both for food webs and for the amount of organic

material reaching bottom waters. How this vertical

redistribution of PP may affect ecosystem function

remains, however, to be quantified.

Acknowledgments

This study received funding from grants 2104-09-063212 and

2104-09-67259 from the Strategic Research Council of Denmark.

Additional support was received from the Danish National

Research Foundation via a grant to the Center for Macroecology

Evolution and Climate, University of Copenhagen, and from the

Department for Bioscience, Aarhus University. We thank

the Department for Bioscience, Aarhus University, for access to

the monitoring data and all the people that have analyzed the

The data did not allow us to differentiate between

adaptation occurring at the population level (i.e., through

changing species composition) and that occurring as

acclimatization at the cellular level. Given, however, the

very different light and nutrient conditions found in the

surface layer and PBL, respectively, it seems likely that

different species may have populated the two environ-

ments. Earlier studies have documented significant vertical

heterogeneity in the distribution of phytoplankton species

in the southern Kattegat (Mouritsen and Richardson 2003).

Further evidence that the populations found in the PBL are

photosynthetically active is found in the seasonal changes

(highest values being found during summer months, when

light intensities are greatest) noted for the P vs. E

parameters derived from populations from the PBL.

Other nitrogen sources—The

fact that we identify such a

clear relationship between the land-based N-loadings origi-

nating from Denmark and the vertical distribution of PP

occurring at six stations in the BSTZ may, initially, seem

surprising, as the BSTZ receives nutrients from several sources

(e.g., advection from surrounding seas, atmospheric deposi-

tion, and land-based runoff from several countries). Nitrogen

budgets for the area (Jørgensen et al. 2013) show that land-

based N-loading only constitute 14% of the total nitrogen

input to the region and 21% of the bioavailable N input. In

addition, only about half of the land-based loadings come

from Denmark. The explanation for the relationship recorded

may possibly be found in the fact that all of the stations in this

study are located close to the Danish coast. As a result, land-

based loading from Germany and Sweden is largely processed

prior to its reaching the sample locations. We attempted to

include the German and Swedish loadings in the regression

analyses (data not shown) but this gave poorer fits than when

we only included the Danish land-based loadings.

Nitrogen entering the BSTZ from adjacent seas consti-

tutes 58% of the bioavailable N (Jørgensen et al. 2013)

introduced to the region. However, these inputs probably

constitute a rather-constant background that leaves the

changes in land-based loading as a likely potential

candidate in generating interannual variability in PP.

Furthermore, some of those inputs may not be reaching

the surface layer during the productive season (see

arguments in Jørgensen et al. 2013).

The strong coupling between local loading and PP means

that local load reductions also can be effective in improving

the marine environment in the area. Previous studies (as

discussed in Duarte 2009) suggest that a return to the

previous condition after load reductions is not to be

expected based on data for phytoplankton biomass.

However, the results in Table 2 and Fig. 6 indicate a linear

response between PP and N-loading in the BSTZ. The

relationship found in this study between PP and N-loading

suggests that eutrophication may, indeed, be reversible and

that it can be abated by reduction in local N-loading, as

phytoplankton PP is the key process starting the cascade of

effects in marine nutrient eutrophication.

Oxygen production in the PBL—The

vertical redistribu-

tion of PP in relation to land-based N-loading demonstrated

Baltic primary production and N-loading

samples over the years; Morten Holtegaard Nielsen for producing

the map of the research area; and the reviewers for valuable

comments.

1689

References

ENDTSEN

, J., K. E. G

USTAFSSON

, J. S

ODERKVIST

AND

J. L. S.

ANSEN

. 2009. Ventilation of bottom water in the North Sea–

Baltic Sea transition zone. J. Mar. Syst.

75:

138–149,

doi:10.1016/j.jmarsys.2008.08.006

OESCH

, D. 2002. Challenges and opportunities for science in

reducing nutrient over-enrichment of coastal ecosystems.

Estuaries

25:

886–900, doi:10.1007/BF02804914

ORUM

, J.,

AND

K. S

AND

-J

ENSEN

. 1996. Is total primary production

in shallow coastal marine waters stimulated by nitrogen

loading? Oikos

76:

406–410, doi:10.2307/3546213

ARSTENSEN

, J., M. S

ANCHEZ

-C

AMACHO

, C. M. D

UARTE

, D.

RAUSE

-J

ENSEN

AND

N. M

ARBA

. 2011. Connecting the dots:

Responses of coastal ecosystems to changing nutrient

concentrations. Environ. Sci. Technol.

45:

9122–9132,

doi:10.1021/es202351y

LOERN

, J. E. 2001. Our evolving conceptual model of the coastal

eutrophication problem. Mar. Ecol. Prog. Ser.

210:

223–253,

doi:10.3354/meps210223

ONLEY

, D. 1999. Biogeochemical nutrient cycles and nutrient

management strategies. Hydrobiologia

410:

87–96,

doi:10.1023/A:1003784504005

ONLEY

, D. J., J. C

ARSTENSEN

, G. Æ

RTEBJERG

, P. B. C

HRISTENSEN

T. D

ALSGAARD

, J. L. S. H

ANSEN

AND

A. B. J

OSEFSON

. 2007.

Long-term changes and impacts of hypoxia in Danish coastal

waters. Ecol. Appl.

17:

165–184, doi:10.1890/05-0766.1

———, S. M

ARKAGER

, J. H. A

NDERSEN

, T. E

LLERMAN

AND

L. M.

VENDSEN

. 2002. Coastal eutrophication and the Danish

National Aquatic Monitoring and Assessment Program.

Estuaries

25:

706–719, doi:10.1007/BF02804910

ULLEN

, J. J. 1982. The deep chlorophyll maximum: Comparing

vertical profiles of chlorophyll

Can. J. Fish.Aquat. Sci.

39:

791–803, doi:10.1139/f82-108

ANISH

TANDARDS

, D. 1986. Vandundersøgelse 2201. Chloro-

phyll

Spectrophotometric measurement in ethanol extracts.

Danish Standards. Water Examination 2201 [References

included].

ORTCH

, Q., N. R

ABALAIS

, R. T

URNER

AND

G. R

OWE

. 1994.

Respiration rates and hypoxia on the Louisiana shelf. Estuar.

Coasts

17:

862–872, doi:10.2307/1352754

UARTE

, C. M. 2009. Coastal eutrophication research: A new

awareness, p. 263–269.

J. H. Andersen and D. J. Conley

[eds.], Eutrophication in coastal ecosystems. Springer.

OFONOFF

, N. P. 1985. Physical properties of seawater: A new

salinity scale and equation of state for seawater. J. Geophys.

Res. Oceans

90:

3332–3342, doi:10.1029/JC090iC02p03332

USTAFSSON

, B. G. 2000. Time-dependent modeling of the Baltic

entrance area. 2. Water and salt exchange of the Baltic Sea.

Estuaries

23:

253–266, doi:10.2307/1352831

ANSEN

, J. L. S.,

AND

J. B

ENDTSEN

. 2013. Parameterisation of

oxygen dynamics in the bottom water of the Baltic Sea–North

Sea transition zone. Mar. Ecol. Prog. Ser.

481:

25–39,

doi:10.3354/meps10220

ØRGENSEN

, B. B.,

AND

K. R

ICHARDSON

. 1996. Eutrophication in

coastal marine ecosystems. Coast. Estuar. Stud.

52:

273.

ØRGENSEN

, L., S. M

ARKAGER

AND

M. M

AAR

. 2013. On the

importance of quantifying bioavailable nitrogen instead of

total nitrogen. Biogeochemistry

117:

1–18.

AAS

, H.,

AND

S. M

ARKAGER

. 1998. Technical guidlines for marine

monitoring. National Environmental Research Institute,

Denmark.

ARLSON

, B., L. E

DLER

, W. G

RANELI

, E. S

AHLSTEN

AND

UYLENSTIERNA

. 1996. Subsurface chlorophyll maxima in

the Skagerrak—processes and plankton community structure.

J. Sea Res.

35:

139–158, doi:10.1016/S1385-1101(96)90742-X

EMP

, M. W.,

AND

W. R. B

OYNTON

. 1984. Spatial and temporal

coupling of nutrient inputs to estuarine primary production:

The role of particulate transport and decomposition. Bull.

Mar. Sci.

35:

522–535.

RAUSE

-J

ENSEN

, D., S. M

ARKAGER

AND

T. D

ALSGAARD

. 2012.

Benthic and pelagic primary production in different nutrient

regimes. Estuar. Coasts

35:

527–545, doi:10.1007/s12237-011-

9443-1

EHRTER

, J. C., M. C. M

URRELL

AND

J. C. K

URTZ

. 2009.

Interactions between freshwater input, light, and phytoplank-

ton dynamics on the Louisiana continental shelf. Cont. Shelf

Res.

29:

1861–1872, doi:10.1016/j.csr.2009.07.001

EWIS

, M. R., J. J. C

ULLEN

AND

T. P

LATT

. 1984. Relationships

between vertical mixing and photoadaptation of phytoplank-

ton: Similarity criteria. Mar. Ecol. Prog. Ser.

15:

141–149,

doi:10.3354/meps015141

UND

-H

ANSEN

, L. C.,

AND OTHERS

. 2008. A consistent high

primary production and chlorophyll-a maximum in a narrow

strait—effects of hydraulic control. J. Mar. Syst.

74:

395–405,

doi:10.1016/j.jmarsys.2008.03.005

ARKAGER

, S. 1993. Light absorption and quantum yield for

growth in five species of marine macroalgae. J. Phycol.

29:

54–63, doi:10.1111/j.1529-8817.1993.tb00279.x

———. 1998. Dark uptake of inorganic

C in oligotrophic

oceanic waters. J. Plankton Res.

20:

1813–1836, doi:10.1093/

plankt/20.9.1813

———, J. C

ARSTENSEN

, D. K

RAUSE

-J

ENSEN

, J. W

INDOLF

AND

IMMERMANN

. 2010. Effects of increased nitrogen load on the

environment in Danish Fjords. National Environmental

Research Institute, Aarhus Univ.

———, W. V

INCENT

AND

E. Y. T

ANG

. 1999. Carbon fixation by

phytoplankton in high Arctic lakes: Implications of low

temperature for photosynthesis. Limnol. Oceanogr.

44:

597–607, doi:10.4319/lo.1999.44.3.0597

———,

AND

W. F. V

INCENT

. 2001. Light absorption by

phytoplankton: Development of a matching parameter for

algal photosynthesis under different spectral regimes. J.

Plankton Res.

23:

1373–1384, doi:10.1093/plankt/23.12.1373

OURITSEN

, L. T.,

AND

K. R

ICHARDSON

. 2003. Vertical microscale

patchiness in nano- and microplankton distributions in a

stratified estuary. J. Plankton Res.

25:

783–797, doi:10.1093/

plankt/25.7.783

URRELL

, M. C., J. G. C

AMPBELL

, J. D. H

AGY

, III,

AND

J. M.

AFFREY

. 2009. Effects of irradiance on benthic and water

column processes in a Gulf of Mexico estuary: Pensacola Bay,

Florida, USA. Estuar. Coast. Shelf Sci.

81:

501–512,

doi:10.1016/j.ecss.2008.12.002

IELSEN

, E. S. 1952. The use of radio-active carbon (

C) for

measuring organic production in the sea. J. Cons. int. Explor.

Mer

18:

117–140, doi:10.1093/icesjms/18.2.117

IELSEN

, S., K. S

AND

-J

ENSEN

, J. B

ORUM

AND

O. G

EERTZ

-H

ANSEN

2002. Depth colonization of eelgrass (Zostera

marina)

and

macroalgae as determined by water transparency in Danish

coastal waters. Estuaries

25:

1025–1032, doi:10.1007/

BF02691349

IXON

, S. W. 1995. Coastal marine eutrophication: A definition,

social causes, and future concerns. Ophelia

41:

199–219.

ETERSEN

, D. L. J.,

AND

M. H

JORTH

. 2010. Marine omrader 2009.

NOVANA. State and trends in the environment and nature

quality. Faglig rapport fra DMU 800 Danmarks Miljøunder-

søgelser, Aarhus Univ., Roskilde. NERI Technical Report.

1690

Lyngsgaard et al.

IMMERMANN

, K., S. M

ARKAGER

AND

K. E. G

USTAFSSON

. 2010.

Streams or open sea? Tracing sources and effects of nutrient

loadings in a shallow estuary with a 3D hydrodynamic–

ecological model. J. Mar. Syst.

82:

111–121, doi:10.1016/

j.jmarsys.2010.04.008

EBB

, W. L., M. N

EWTON

AND

D. S

TARR

. 1974. Carbon dioxide

exchange of

Alnus rubra.

Oecologia

17:

281–291, doi:10.1007/

BF00345747

INDOLF

, J., G. B

LICHER

-M

ATHIESEN

, J. C

ARSTENSEN

AND

RONVANG

. 2012. Changes in nitrogen loads to estuaries

following implementation of governmental action plans in

Denmark: A paired catchment and estuary approach for

analysing regional responses. Environ. Sci. Policy

24:

24–33,

doi:10.1016/j.envsci.2012.08.009

———, H. T

HODSEN

, L. T

ROLDBORG

, S. E. L

ARSEN

, J. B

ØGESTRAND

N. B. O

VESEN

AND

B. K

RONVANG

. 2011. A distributed modeling

system for simulation of monthly runoff and nitrogen sources,

loads and sinks for ungauged catchments in Denmark. J.

Environ. Monit.

13:

2645–2658, doi:10.1039/c1em10139k

ICHARDSON

, K., J. B

EARDALL

AND

J. A. R

AVEN

. 1983. Adaptation

of unicellular algae to irradiance: An analysis of strategies.

New Phytol.

93:

157–191, doi:10.1111/j.1469-8137.1983.

tb03422.x

———,

AND

A. C

HRISTOFFERSEN

. 1991. Seasonal distribution and

production of phytoplankton in the Southern Kattegat. Mar.

Ecol. Prog. Ser.

78:

217–227, doi:10.3354/meps078217

———, B. R

ASMUSSEN

, T. B

UNK

AND

L. T. M

OURITSEN

. 2003.

Multiple subsurface phytoplankton blooms occurring simul-

taneously in the Skagerrak. J. Plankton Res.

25:

799–813,

doi:10.1093/plankt/25.7.799

———, A. W. V

ISSER

AND

F. B. P

EDERSEN

. 2000. Subsurface

phytoplankton blooms fuel pelagic production in the North Sea.

J. Plankton Res.

22:

1663–1671, doi:10.1093/plankt/22.9.1663

AVCHUK

, O. P. 2005. Resolving the Baltic Sea into seven

subbasins: N and P budgets for 1991–1999. J. Mar. Syst.

56:

1–15, doi:10.1016/j.jmarsys.2004.08.005

IEGEL

, D. A.,

AND

A. F. M

ICHAELS

. 1996. Quantification of non-

algal light attenuation in the Sargasso Sea: Implications for

biogeochemistry and remote sensing. Deep-Sea Res. II

43:

321–345, doi:10.1016/0967-0645(96)00088-4

MITH

, V. 2003. Eutrophication of freshwater and coastal marine

ecosystems a global problem. Environ. Sci. Pollut. Res.

10:

126–139, doi:10.1065/espr2002.12.142

TRICKLAND

, J. D. H.,

AND

T. R. P

ARSONS

. 1972. A practical handbook

of seawater analysis. Bull. Fish. Res. Bd. Can 167.

TROM

, S. L., E. L. M

ACRI

AND

K. A. F

REDRICKSON

. 2010. Light

limitation of summer primary production in the coastal Gulf

of Alaska: Physiological and environmental causes. Mar.

Ecol. Prog. Ser.

402:

45–57, doi:10.3354/meps08456

Associate editor: Anthony W. D. Larkum

Received: 07 November 2013

Accepted: 23 May 2014

Amended: 28 May 2014